Travel using managed lanes: An application of a stated choice model for Houston, Texas

Abstract

Managed lane (ML) travel adds flexibility, but also complexity, to travel choices. Stated choice models (SCMs) are often used for modeling complex transportation choices such as these in an effort to predict demand for these travel options. The design methods for SCMs have evolved from simple orthogonal designs to more sophisticated designs such as D-efficient design that can increase efficiency in estimation. We used three different survey design strategies to produce the stated preference portion of surveys, which were used to elicit travel choices for a sample of Houston travelers. Apart from the D-efficient design we also used random and adaptive random designs to generate attribute levels. There were observable differences in choice behavior depending on what design strategy was used. These differences appear to influence estimates of the value of travel time savings (VTTS) obtained from the random parameter logit (RPL) models estimated using these data. This, in turn, would greatly impact the percentage of travelers predicted to use the MLs. The adaptive random strategy was superior to the other design methods in several categories, and it had similar efficiency to the D-efficient design. However, the mean of VTTS estimate obtained from a D-efficient design was closer to what is typically found in the literature. The difference was considerable and could greatly influence traffic and revenue estimates for the MLs, illustrating the importance of the survey design strategy.